Orange Forum • View topic - accuracy3.py shows different result from doc

accuracy3.py shows different result from doc

A place to ask questions about methods in Orange and how they are used and other general support.

accuracy3.py shows different result from doc

Postby Guest » Tue Jan 11, 2005 22:32

Hello,
The online doc http://www.ailab.si/orange/doc/ofb/c_performance.htm
said the following result by running doc/ofb/accuracy3.py

The output of this testing is:
Classification accuracies:
bayes 0.903448275862
tree 0.965517241379

But on my machine accuracy3.py shows quite different result:
python accuracy3.py
Classification accuracies:
bayes 0.935632183908
tree 0.901149425287

The result of bayes is higher and that of tree is lower in my case. I'm using orange 0.99b with C4.5 plugin installed. Why?

Postby Guest » Tue Jan 11, 2005 23:33

I then download and install the latest cvs version 2005-01-11 and find the same problem is still there. Moreover, I found the output of accuracy4.py has problem too.

python accuracy4.py
1: [0.89312977099236646, 0.80152671755725191]
2: [0.89312977099236646, 0.80152671755725191]
3: [0.89312977099236646, 0.80152671755725191]
4: [0.89312977099236646, 0.80152671755725191]
5: [0.89312977099236646, 0.80152671755725191]
6: [0.89312977099236646, 0.80152671755725191]
7: [0.89312977099236646, 0.80152671755725191]
8: [0.89312977099236646, 0.80152671755725191]
9: [0.89312977099236646, 0.80152671755725191]
10: [0.89312977099236646, 0.80152671755725191]
Classification accuracies:
bayes 0.893129770992
tree 0.801526717557

All the 10-fold CV yield the same result, which is also different from the doc.


Return to Questions & Support



cron